
import pandas as pd
import numpy as np


#IMPORTING THE DATASET
tvshows_df = pd.read_csv('tv_shows.csv')
tvshows_df


#DATA CLEANING
new_df=tvshows_df[tvshows_df.IMDb.notna()]
new_df=new_df[new_df.Age.notna()]
new_df=new_df[new_df['Rotten Tomatoes'].notna()]
new_df.reset_index(inplace=True)
new_df=new_df.drop(columns=['index','Unnamed: 0','type'])


#CONVERTING THE NUMERICAL COLUMN TO Decipherable type
IMDb_rating=np.array(list(new_df.IMDb))
RT_rating=np.array(np.char.strip(np.array(list(new_df["Rotten Tomatoes"])),'%'),dtype='float64')
new_df['Rotten Tomatoes']=RT_rating
#cleaned data frame
new_df


#finding the minimum and maximum IMDb ratings
new_df[new_df.IMDb==new_df.IMDb.min()]


new_df[new_df.IMDb==new_df.IMDb.max()]


import seaborn as sns
import matplotlib
import matplotlib.pyplot as plt
%matplotlib inline

sns.set_style('darkgrid')
matplotlib.rcParams['font.size'] = 15
matplotlib.rcParams['figure.figsize'] = (9, 5)
matplotlib.rcParams['figure.facecolor'] = '#00000000'


#PLOTTING HISTOGRAM
fig, ax = plt.subplots(figsize=(12,9))
ax.hist(new_df.Year,bins=5, density=False)
plt.xlabel("Year")
plt.ylabel("Rotten Tomatoes")
plt.title("Year")
for rect in ax.patches:
    height = rect.get_height()
    ax.annotate(f'{int(height)}', xy=(rect.get_x()+rect.get_width()/2, height),xytext=(0, 5), textcoords='offset points', ha='center', va='bottom') 
plt.show()


new_df.describe()


#rating comparison using the boxplot
fig,ax = plt.subplots()
bp_data=[IMDb_rating*10,RT_rating]
ax.boxplot(bp_data)
ax.set_xticklabels(["IMDb","Rotten tomatoes"])
plt.ylabel("Relative Ratings(Scale 0-100)")
plt.title("Ratings ")
plt.show()


#Z-Score Computation Function
def z_score(df):
    df_std = df.copy()
    for column in df_std.columns:
        df_std[column] = (df_std[column] - df_std[column].mean()) / df_std[column].std()
    return df_std


new_df_numeric = new_df[["Year","IMDb","Rotten Tomatoes"]]
df_normalized = z_score(new_df_numeric)


#year normalisation
print("Normalized Curve For Year:-")
print("Mean :",df_normalized.Year.mean())
print("Standard Deviation :",df_normalized.Year.std())

Normalized Curve For Year:-
Mean : 1.240480447804325e-14
Standard Deviation : 1.0000000000000013


#IMDb normaisation
print("Normalized Curve For IMDb Rating:")
print("Mean :",df_normalized.IMDb.mean())
print("Standard Deviation :",df_normalized.IMDb.std())

Normalized Curve For IMDb Rating:
Mean : 6.537794532980621e-15
Standard Deviation : 0.9999999999999976


#rotten tomatoes normalisation
print("Normalized Curve For Rotten Tomatoes Rating:")
print("Mean :",df_normalized['Rotten Tomatoes'].mean())
print("Standard Deviation :",df_normalized['Rotten Tomatoes'].std())

Normalized Curve For Rotten Tomatoes Rating:
Mean : -9.671223125359849e-17
Standard Deviation : 0.9999999999999997


new_df.corr()


subset_df=new_df[['IMDb','Rotten Tomatoes']]
subset_df.corr()
sns.regplot(x="IMDb", y="Rotten Tomatoes", data=subset_df);


#IMDb rating using coorelation analysis (min and max without outliers)
IMDb_rating=np.array(new_df.IMDb)
imdb_iqr=np.percentile(IMDb_rating,75)-np.percentile(IMDb_rating,25)
min_IMDb_no_outliers=max(np.percentile(IMDb_rating,25)-1.5*imdb_iqr,IMDb_rating.min())
max_IMDb_no_outliers=min(np.percentile(IMDb_rating,75)+1.5*imdb_iqr,IMDb_rating.max())
print("Minimum IMDb rating without outliers is ",min_IMDb_no_outliers)
print("Maximum IMDb rating without outliers is ",max_IMDb_no_outliers)

Minimum IMDb rating without outliers is  5.45
Maximum IMDb rating without outliers is  9.5


#Rotten tomatoes rating using correlation comparison (min and max without outliers)
RT_rating=np.array(new_df['Rotten Tomatoes'])
rt_iqr=np.percentile(RT_rating,75)-np.percentile(RT_rating,25)
min_RT_no_outliers=max(np.percentile(RT_rating,25)-1.5*rt_iqr,RT_rating.min())
max_RT_no_outliers=min(np.percentile(RT_rating,75)+1.5*rt_iqr,RT_rating.max())
print("Minimum Rotten Tomatoes rating without outliers is ",min_RT_no_outliers)
print("Maximum Rotten Tomatoes rating without outliers is ",max_RT_no_outliers)

Minimum Rotten Tomatoes rating without outliers is  28.0
Maximum Rotten Tomatoes rating without outliers is  100.0


#removing the outliers
subset_df=new_df[['IMDb','Rotten Tomatoes']]
subset_df=subset_df[subset_df.IMDb>=min_IMDb_no_outliers]
subset_df=subset_df[subset_df.IMDb<=max_IMDb_no_outliers]
subset_df=subset_df[subset_df['Rotten Tomatoes']>=min_RT_no_outliers]
subset_df=subset_df[subset_df['Rotten Tomatoes']<=max_RT_no_outliers]
subset_df


#plotting graph for the  without outliers
sns.regplot(x="IMDb", y="Rotten Tomatoes", data=subset_df);


subset_df.corr()


import numpy as np
import pandas as pd
tv_shows = pd.read_csv('tv_shows.csv')
tv_shows


#data cleaning
tv_shows.drop_duplicates(subset='Title',keep='first',inplace=True)
tv_shows['Rotten Tomatoes'] = tv_shows['Rotten Tomatoes'].fillna('0%')
tv_shows['Rotten Tomatoes'] = tv_shows['Rotten Tomatoes'].apply(lambda x : x.rstrip('%'))
tv_shows['Rotten Tomatoes'] = pd.to_numeric(tv_shows['Rotten Tomatoes'])

tv_shows['IMDb'] = tv_shows['IMDb'].fillna(0)
tv_shows['IMDb'] = tv_shows['IMDb']*10
tv_shows['IMDb'] = tv_shows['IMDb'].astype('int')


tv_shows_long=pd.melt(tv_shows[['Title','Netflix','Hulu','Disney+','Prime Video']],id_vars=['Title'],var_name='StreamingOn', value_name='Present')
tv_shows_long = tv_shows_long[tv_shows_long['Present'] == 1]
tv_shows_long.drop(columns=['Present'],inplace=True)


#merging datasets
tv_shows_combined = tv_shows_long.merge(tv_shows, on='Title', how='inner')


tv_shows_combined.drop(columns = ['Unnamed: 0','Netflix','Hulu', 'Prime Video', 'Disney+', 'type'], inplace=True)


#Subsetting.The datasets with IMDB ratings/ Rotten Tomatoes ratings above 0 needs to be considered for plotting.
tv_shows_both_ratings = tv_shows_combined[(tv_shows_combined.IMDb > 0) & tv_shows_combined['Rotten Tomatoes'] > 0]
tv_shows_both_ratings


tv_shows_combined.groupby('StreamingOn').Title.count().plot(kind='bar')

<matplotlib.axes._subplots.AxesSubplot at 0x29176514348>


#Correlation matrix
import seaborn as sns
import matplotlib.pyplot as plt
corrmat = df.corr()
fig = plt.figure(figsize = (12, 9))

sns.heatmap(corrmat, vmax = .8, square = True, annot = True)
plt.show()


#k clustering
import pandas as pd
DF = pd.read_csv('tv_shows.csv', encoding = 'utf-8')
DF


import numpy as np
from sklearn import preprocessing
DF_ARRAY = np.array(DF.iloc[:,6:11])          # Getting only the numeric features from the dataset
DF_NORM  = preprocessing.normalize(DF_ARRAY) # Normalizing the data
DF_NORM

array([[0.70710678, 0.        , 0.        , 0.        , 0.70710678],
       [0.70710678, 0.        , 0.        , 0.        , 0.70710678],
       [0.70710678, 0.        , 0.        , 0.        , 0.70710678],
       ...,
       [0.        , 0.        , 0.        , 0.70710678, 0.70710678],
       [0.        , 0.        , 0.        , 0.70710678, 0.70710678],
       [0.        , 0.        , 0.        , 0.70710678, 0.70710678]])


from sklearn.cluster import KMeans

# Creating our Model
kmeans = KMeans(n_clusters = 3)


# Training our model
kmeans.fit(DF_NORM)

KMeans(n_clusters=3)


import matplotlib.pyplot as plt

# Amount of values to be tested for K
Ks = range(6, 11)

# List to hold on the metrics for each value of K
results = []

# Executing the loop
for K in Ks:
    
    model = KMeans(n_clusters = K)
    model.fit(DF_NORM)
    
    results.append(model.inertia_)

# Plotting the final result
plt.plot(Ks, results, 'o-')
plt.xlabel("Values of K")
plt.ylabel("tv_shows")
plt.show()


from sklearn.cluster import KMeans

# Creating our Model
kmeans = KMeans(n_clusters = 4)

# Training our model
kmeans.fit(DF_NORM)

# You can see the labels (clusters) assigned for each data point with the function labels_
kmeans.labels_

# Assigning the labels to the initial dataset
DF['cluster'] = kmeans.labels_


from matplotlib import pylab
from pylab import *


from sklearn.decomposition import PCA
import pylab as pl

# Reducing data dimensions 
PCA_ = PCA(n_components = 2).fit(DF_ARRAY)

# Applying the PCA
PCA_2 = PCA_.transform(DF_ARRAY)

#--------------------------
# Plotting the cluster
#--------------------------

# Plot size
pylab.rcParams['figure.figsize'] = (8.0, 8.0)

# Plotting each point individually depending on their cluster
for i in range(0, PCA_2.shape[0]):
    
    # If the 'i' data point be in cluster 0, it will be plotted as the formatting inside the if functions
    # And so on...
    if kmeans.labels_[i] == 0:
        CLUSTER_01 = pl.scatter(PCA_2[i,0], PCA_2[i,1], c ='r', marker = 'o', s = 120)
        
    elif kmeans.labels_[i] == 1:
        CLUSTER_02 = pl.scatter(PCA_2[i,0], PCA_2[i,1], c ='g', marker = 'o', s = 120)

    elif kmeans.labels_[i] == 2:
        CLUSTER_03 = pl.scatter(PCA_2[i,0], PCA_2[i,1], c ='b', marker = 'o', s = 120)

    elif kmeans.labels_[i] == 3:
        CLUSTER_04 = pl.scatter(PCA_2[i,0], PCA_2[i,1], c ='y', marker = 'o', s = 120)
        
        
        # Formatting the Plot
        pl.legend([CLUSTER_01, CLUSTER_02, CLUSTER_03, CLUSTER_04],
                  ['Cluster 01', 'Cluster 02', 'Cluster 03', 'Cluster 04'])
        
        pl.title('TV_SHOWS')
        
pl.show()


import pandas as pd
import numpy as np

import missingno as msno

import seaborn as sns
import matplotlib.pyplot as plt
import plotly.express as px
import plotly.graph_objects as go

import warnings 
warnings.filterwarnings('ignore')

%matplotlib inline


df=pd.read_csv("tv_shows.csv")
df.head()


# Visualize missing values as a matrix
msno.matrix(df)

<matplotlib.axes._subplots.AxesSubplot at 0x2917f2d56c8>


msno.bar(df)

<matplotlib.axes._subplots.AxesSubplot at 0x2917f332488>


for i in df.columns:
    null_rate = df[i].isnull().sum() / len(df)*100
    if null_rate > 0 :
        print( "{}'s null rate: {}%".format(i, round(null_rate, 2)))

Age's null rate: 43.59%
IMDb's null rate: 20.69%
Rotten Tomatoes's null rate: 81.98%


#wordcloud
from wordcloud import WordCloud, STOPWORDS

text = ' '.join(df['Title'])

plt.rcParams['figure.figsize'] = (6,12)
wordcloud = WordCloud(background_color = 'black',colormap='vlag', width = 1200,  height = 1200, max_words = 121).generate(text)
plt.imshow(wordcloud)
plt.axis('off')
plt.show()


col_names_online_media = ['Netflix', 'Prime Video', 'Disney+', 'Hulu']
col_names_rating_sites = ['IMDb', 'Rotten Tomatoes']


pallete_dict = {
    'Netflix': ['#A9A9A9', '#E50914'],
    'Prime Video': ['#A9A9A9', '#00A8E1'],
    'Hulu' : ['#A9A9A9', '#3DBB3D'],
    'Disney+' : ['#A9A9A9', '#113CCF']
}


fig = plt.figure(figsize=[20,10])
sns.set_style('dark')

for col_name, num in zip(col_names_online_media, 
                         range(1,len(col_names_online_media) + 1)):
    ax = fig.add_subplot(2, 2, num)
    sns.scatterplot(x="Year", y="Title", 
                    palette = pallete_dict[col_name],
                    hue=col_name, 
                    data=df.sort_values(by=[col_name, 'Title']), 
                    ax=ax)
    sns.despine
    ax.set_title('TV Shows on ' + col_name, fontsize=25)
    handles, labels = ax.get_legend_handles_labels()
    ax.legend(loc='upper left', 
              frameon=None,
              edgecolor='black',
              fontsize=15,
              framealpha=0.2,
              handles=[handles[2]],
              labels=['On ' + col_name])
    ax.set_xlim(1900, 2022)
    ax.set(yticklabels=[])
    ax.set(ylabel=None, xlabel=None)
    
fig.tight_layout()

for ax in fig.get_axes():
    ax.label_outer()

plt.show()


dict_subscribers_Millions = {'Netflix':183, 'Amazon Prime Video': 75, 'Disney+': 54.5, 'Hulu': 30 }


import pandas as pd
import matplotlib.pyplot as plt
import seaborn as sns

import re
import string

fig = plt.figure(figsize=[10,8])
sns.set_style('darkgrid')

ax = sns.barplot(x=list(dict_subscribers_Millions.keys()), y=list(dict_subscribers_Millions.values()),
            palette = ['#E50914', '#00A8E1', '#113CCF' , '#3DBB3D'] 
           )
plt.ylabel('Subscribers in Millions')
plt.savefig('demo9.png', transparent=True)

	Unnamed: 0	Title	Year	Age	IMDb	Rotten Tomatoes	Netflix	Hulu	Prime Video	Disney+	type
0	0	Breaking Bad	2008	18+	9.5	96%	1	0	0	0	1
1	1	Stranger Things	2016	16+	8.8	93%	1	0	0	0	1
2	2	Money Heist	2017	18+	8.4	91%	1	0	0	0	1
3	3	Sherlock	2010	16+	9.1	78%	1	0	0	0	1
4	4	Better Call Saul	2015	18+	8.7	97%	1	0	0	0	1
...	...	...	...	...	...	...	...	...	...	...	...
5606	5606	Tut's Treasures: Hidden Secrets	2018	NaN	NaN	NaN	0	0	0	1	1
5607	5607	Paradise Islands	2017	NaN	NaN	NaN	0	0	0	1	1
5608	5608	Wild Russia	2018	NaN	NaN	NaN	0	0	0	1	1
5609	5609	Love & Vets	2017	NaN	NaN	NaN	0	0	0	1	1
5610	5610	United States of Animals	2016	NaN	NaN	NaN	0	0	0	1	1

	Title	Year	Age	IMDb	Rotten Tomatoes	Netflix	Hulu	Prime Video	Disney+
0	Breaking Bad	2008	18+	9.5	96.0	1	0	0	0
1	Stranger Things	2016	16+	8.8	93.0	1	0	0	0
2	Money Heist	2017	18+	8.4	91.0	1	0	0	0
3	Sherlock	2010	16+	9.1	78.0	1	0	0	0
4	Better Call Saul	2015	18+	8.7	97.0	1	0	0	0
...	...	...	...	...	...	...	...	...	...
926	Diary of a Future President	2020	7+	5.5	100.0	0	0	0	1
927	Encore!	2019	7+	7.4	68.0	0	0	0	1
928	Spider-Man Unlimited	1999	7+	6.5	50.0	0	0	0	1
929	The Super Hero Squad Show	2009	7+	6.1	50.0	0	0	0	1
930	Marvel's Hero Project	2019	7+	6.1	92.0	0	0	0	1

	Year	IMDb	Rotten Tomatoes	Netflix	Hulu	Prime Video	Disney+
count	931.000000	931.000000	931.000000	931.000000	931.000000	931.000000	931.000000
mean	2013.093448	7.534801	77.823845	0.446831	0.375940	0.227712	0.024705
std	7.606510	0.952062	20.369783	0.497432	0.484625	0.419581	0.155307
min	1959.000000	1.700000	6.000000	0.000000	0.000000	0.000000	0.000000
25%	2011.000000	7.100000	67.000000	0.000000	0.000000	0.000000	0.000000
50%	2015.000000	7.700000	84.000000	0.000000	0.000000	0.000000	0.000000
75%	2018.000000	8.200000	93.000000	1.000000	1.000000	0.000000	0.000000
max	2020.000000	9.500000	100.000000	1.000000	1.000000	1.000000	1.000000

	Year	IMDb	Rotten Tomatoes	Netflix	Hulu	Prime Video	Disney+
Year	1.000000	-0.233711	-0.063455	0.183902	-0.216642	-0.204441	-0.016520
IMDb	-0.233711	1.000000	0.481183	0.008906	-0.005082	0.082427	-0.019638
Rotten Tomatoes	-0.063455	0.481183	1.000000	0.021997	0.017935	-0.016312	0.026189
Netflix	0.183902	0.008906	0.021997	1.000000	-0.568219	-0.405600	-0.143042
Hulu	-0.216642	-0.005082	0.017935	-0.568219	1.000000	-0.252236	-0.080669
Prime Video	-0.204441	0.082427	-0.016312	-0.405600	-0.252236	1.000000	-0.086422
Disney+	-0.016520	-0.019638	0.026189	-0.143042	-0.080669	-0.086422	1.000000

	IMDb	Rotten Tomatoes
0	9.5	96.0
1	8.8	93.0
2	8.4	91.0
3	9.1	78.0
4	8.7	97.0
...	...	...
926	5.5	100.0
927	7.4	68.0
928	6.5	50.0
929	6.1	50.0
930	6.1	92.0

DATASET¶

DATA CLEANING¶

HISTOGRAM¶

BOXPLOT¶

PIECHART¶

Z-SCORE AND NORMALISATION¶

IMDB VS ROTTEN TOMATOES¶

CORRELATION ANALYSIS¶

REMOVING THE OUTLIERS¶

COMBINED DATASET BARGRAPH¶

CORRELATION MATRIX¶

K-CLUSTERING¶

ANALYSIS AND VISUALIZATION ON THE MISSING DATA¶

Top 10 imdb movies and shows¶

WORDCLOUD¶

SCATTERPLOT¶

SUBSCRIBERS COUNT¶

Total number of TV Shows on each of these online streaming websites:¶

Rotten Tomatoes vs the online streaming channels¶

IMDB vs the online streaming channels¶

Age vs the online streaming channels¶

TV SHOWS AND MOVIES PLATFORM-to find which platform has more movies¶